In Section 5 of his important paper, “Truth, Rationality, and the Growth of Knowledge,”’ Karl Popper discusses the requirements for knowledge to grow. According to Popper, a scientist often finds himself in the following problem-situation: A theory T that he has been using so far, while it can explain some facts, cannot explain others, and is falsified by still other facts. Consequently, the scientist desires to come up with a theory T‘ that will explain everything that T could and at (...) least some things that T could not, especially those facts which falsified T. Would not such a success, if we could emerge with a T’, be enough? Not so, claims Popper. Given that our aim is more and more interesting truth, “the logic of the general problem-situation” demands that T‘ satisfy other requirements as well. If these added requirements are not met, our aim of approach- ing truth would be vitiated.2 These additional requirements are three in all. First, that every new theory proposed, such as T‘, should proceed from some “simple, new, and powerful, unifying idea.” Second, that every new theory should be independently testable, and third, each such theory should pass some new, and severe, tests.3 Popper believes that the third requirement is crucial for the growth of knowledge, and he discusses it in considerable detail. My primary aim in this paper is to show that Popper’s arguments in support of his third requirement do not stand up. I shall argue that if knowledge can grow at all on Popper’s other assumptions then it can grow even if every new theory subsequent to the initial theory should violate Popper’s third requirement. In showing how knowledge can grow even though the third requirement is never satisfied, I use assumptions and methods which Popper allows himself. Hence, I leave open the question of the legitimacy of these assumptions and methods which have, in fact, been severely criticized by Adolf Grunbaum, David Miller, Pave1 Tichy, and others. (shrink)